Open SourceRAG33k
FAISS
Meta's battle-tested library for billion-scale similarity search.
FAISS (Facebook AI Similarity Search) is a library from Meta for efficient similarity search and clustering of dense vectors. It implements many indexing algorithms (IVF, HNSW, PQ, OPQ) with optional GPU acceleration, scaling to billions of vectors that may not fit in RAM. It is the de facto low-level engine that powers vector search inside countless RAG stacks and higher-level vector databases.
Repository
facebookresearch/faissLanguage
C++What you'd build with it
- Building an in-process, high-throughput nearest-neighbor index for a RAG retrieval layer without running a separate database
- Running GPU-accelerated similarity search over very large embedding sets where latency and memory footprint matter
- Benchmarking and tuning index types (IVF, HNSW, product quantization) to trade recall against speed and memory
Tags
vector-searchannembeddingsgpu